Relying on your gut, when it comes to developing marketing strategy, usually leads to more Rolaids than revenue. Developing marketing strategy based on your customers’ preferences and perceptions is always a better bet.
This is the second article of a two-part series on conducting “do it yourself” marketing research.
February’s column (Supply-demand decisions need insight) covered the first two steps: planning and designing a marketing research study. This installment addresses the final two steps: collecting and analyzing.
Collecting: After planning and designing the research project you’ll be ready to conduct the study. But before you begin customer interviews or e-mailing an online survey link, test the instrument and collection process for glitches.
There’s nothing worse than collecting all of the data, reaching your sample goal (the number of respondents you were aiming for) and finding out there was a problem in the collection process that renders your data useless.
Conduct a short test to make sure the survey design and collection methodology are free of errors. Be sure the wording (and sequence) of your questions don’t bias responses. Respondents also can unintentionally cause errors, respondent confusion and fatigue are common.
Examples: if a respondent is confused by a question, they may not be answering the question you intended. That means the data has no validity. Also, if the interview or survey is too long, the respondent may become burned out and the accuracy and completion rates wane.
Be sure the wording of all your questions is simple and clear, and that your questionnaire is not too long — best practices suggest five-minutes max for an online survey and not longer than 15 minutes for a phone interview.
Working with an experienced third party can minimize collection errors; however, as the client, you should review test results to make sure everything is in order before burning through your customer database. Analyzing: You’ll begin this final step by “scrubbing” the data (correcting entry errors). After errors have been omitted, you’ll format, code and tabulate the data.
When possible, it’s best to enter answers directly into a database when conducting interviews. I recommend formatting your research data in a standard spreadsheet such as Excel. Multiple-choice and ranking answers can be coded (using numbers) to speed up entry and tabulation.
By entering question numbers in the column header and responses by row, you can tabulate data by sorting and grouping answers to each question…this process will surface dominate themes. Calculate response groupings to get percentages for each question. You are looking for data (values) that are statistically relevant. For example, if 67 percent of your customer base indicates they prefer product attribute “A” over “B,” you can be relatively confident in featuring (or providing) attribute “A,” knowing it has greater consumer appeal.
A common analysis technique is to cross tabulate the data sets, which is achieved when you combine two or more sets of data, e.g. compare respondent demographic variables (independent) to how a “value” question was answered (dependent variable). You gain a much better understanding of the market when cross tabulating independent to dependent variables — income bracket to preferred attribute, respectively.
Start with a simple project — a basic data set is far better than no intelligence at all. It will take some planning and design before you can conduct and analyze a marketing research study, but the up-front investment will pay dividends down the road. Before making a high-impact business decision, verify the best path to take before embarking on the journey. My favorite quote is “test before you invest,” probably because it is mine.
Andrew Ballard is president of Marketing Solutions, an agency specializing in growth strategies. For more information, call 425-337-1100 or go to www.mktg-solutions.com.